For more details on the courses, please refer to the Course Catalog
| Code | Course Title | Credit | Learning Time | Division | Degree | Grade | Note | Language | Availability |
|---|---|---|---|---|---|---|---|---|---|
| AIM5028 | SW-HW Integrated Design | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| SW-HW Integrated Design Methodology covers SW and HW integrated desgin methods to design the efficient Artificial Intelligence (AI) system for various applications. Optimum partitioning between SW and HW is needed considering the data processing speed, power consumption, and complexity and optimum performance can be achieved. This course covers AI SW design methodology, AI HW design methodology, and AI SW-HW design methodology. | |||||||||
| AIM5029 | AI Colloquium | 1 | 2 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Thisclassprovidesbroadknowledgeaboutmanyfieldsofinformationtechnology.VarioussubjectsareselectedwhicharecurrentlyhotissuesinArtificial Intelligence andinvitedtalksaregivenabouttheselectedsubjects. | |||||||||
| AIM5033 | Multi-modal Learning | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Real-word data usually contain different types of modalities. For example, images can be associated with short descriptions and in a document, some ideas are illustrated by an image even though most of the ideas are delivered by texts. This graduate course will discuss multi-modal learning techniques which integrate different types of data in a machine learning method. | |||||||||
| AIM5035 | Explainable AI | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Recently, deep neural networks are demonstrating superb prediction accuracy, but their complex architectures make it hard to explain its prediction results. In this course, we aim to cover the recent efforts on interpreting the complex decision process of deep neural networks. First, we cover the visualizations of the learned representations of neural networks, then consider the saliency-map based interpretation methods (e.g., Grad-CAM and LRP). We will also cover black-box interpretation methods (e.g., LIME, SHAP) and look at the robustness of such methds. Moreover, we will look at important applications that requires interpretability and carry out term-projects. | |||||||||
| AIM5036 | Deep Generative Model | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | Korean | Yes | |
| We cover the topics on deep generative models, which are attracting lots of attention. First part include autoregressive models, variational autoencoder (VAE), normalized flow, and the second half will be dedicated to generative adversarial networks (GAN) and its variations such as Wasserstein GAN, conditional GAN, and Cycle GAN, et. We will also cover various application areas that utilize generative models. | |||||||||
| AIM5038 | Context-aware Learning | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Context-awareness refers to the idea that a computer (or any programmable device) can sense and react based on their environment. The emphasis is on making a machine appropriately react to the dynamic environment that the machine faces. This graduate course will discuss context-aware learning methods that allow intelligent devices to be aware of their contexts and react based on appropriate decision making. | |||||||||
| AIM5039 | Intelligent Storytelling | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| In this course students will learn theories on interactive storytelling, computational models of interactive narrative based on AI technology, and practice/demonstration/exercise on interactive storytelling systems and authoring tools. After taking the class, students should be able to apply theoretical narrative models, AI techniques, and authoring tools for building interactive narrative systems. In addition, they are expected to propose and evaluate research ideas in interactive storytelling systems | |||||||||
| AIM5040 | Unsupervised learning | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| We cover the basic and advanced topics on unsupervised learning, which directy learns from unlabeled data. The basic topics will cover K-means clustering, principal component analysis (PCA), independent component analysis (ICA), and expectation-maximization (EM), hidden Markov models (HMM). The more advanced topics include restricted Boltzman machine (RBM) and deep Boltzmann machine (DBM). | |||||||||
| AIM5041 | Affective Computing | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| The primary goal of the course is to understand computational models of emotions. The course investigates AI-based technologies and algorithms. The learning contents include definition of emotion, emotion recognition using machine learning algorithms, cognition-based emotion processing model, and multimodal emotion expressions. After taking the course, students are expected to apply the models in different applications. | |||||||||
| AIM5042 | Game AI | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Artificial intelligence is one of the essential components of a computer game, and computer games can be referred as testbeds for human-level intelligence of computers. This lecture introduces various game AI techniques including state machines, decision making algorithms for both realism and entertaining experiences, path finding, and strategy making. | |||||||||
| AIM5043 | AI Accelerator | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Current artificial intelligence hardware architecture is based on the artificial deep neural network model, but requires more innovative structures similar to the human brain and nervous system. Based on the conventional AI accelerator architecture optimized for matrix computation, we explore a variety of next-generation architectures that encompass analog-based neural network circuitry, processing-in-memory, and beyond the von Neumann architecture. | |||||||||
| AIM5044 | Neuromorphic Processor | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| The purpose of the Neuromorphic Processor is to learn a variety of techniques for designing neuromorphic systems mimicking the structure and behavior of a biological brain. After understanding the structure and behavior of a biological brain, the students will learn about the basic properties of memristive devices and CMOS circuits as important tools to implement neuromorphic systems. The lecture also deals with the architectures and operation principles of current neuromorphic processors, and learns how to design low-power high-performance neuromorphic processors. | |||||||||
| AIM5045 | Edge Computing | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| In this course we will study the extension of cloud computing to today’s edge computing and learn how they can be leveraged in a combined edge-cloud- environment. We will begin with a review of current cloud computing environment and the structure of data center networking. We will then learn about the definition of edge computing and the complementary nature of cloud computing and edge computing. It also covers the difference between device edge and cloud edge depending on computing over edge devices and cloud. We will study edge architecture, coordination of cloud services, Apache Edgent, geo-distributed computing, machine learning in edge computing and algorithms for distributed big data analytics. | |||||||||
| AIM5049 | AI Healthcare System | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Modern healthcare system consists of many interlinked complex components. Managing them properly is a key issue in effective healthcare delivery. This course deals with components in the healthcare system and how to manage them using artificial intelligence-based operations management tools. Major topics include design of healthcare capacity, optimization of healthcare facility locations, supply chain management of blood, healthcare information system, and organ distribution models. | |||||||||
| AIM5050 | Medical Image Analysis | 3 | 6 | Major | Master/Doctor | Artificial Intelligence | - | No | |
| Modern healthcare involves increased use of medical imaging and thus it is important to analyze medical imaging data for better healthcare. This course explains various medical imaging modalities and deals with how to extract clinically relevant information from medical imaging using image processing/computer vision techniques. Major topics include principles of magnetic resonance imaging and computed tomography, filter-based feature extraction, and neural network-based feature extraction. | |||||||||